In brief
Ukraine’s government has approved a national text corpus for training and testing large language models. Eligible Ukrainian developers will receive free access under licensing and security rules.
Ukraine’s Cabinet of Ministers has adopted a resolution to launch a national Ukrainian-language data corpus for training and testing large language models, RBC-Ukraine reported, citing the resolution. The report did not specify an adoption or launch date or establish that the platform was already operating.
The Ministry of Digital Transformation will oversee the project, while the state enterprise Diia will handle technical content, structuring and cybersecurity. The corpus is intended to combine anonymised structured and unstructured texts, including government datasets, literature, scientific works and archival materials.
Free access under licence will be limited to Ukrainian legal entities and individual entrepreneurs with demonstrated experience in developing or testing large language models. They must use the material exclusively for those purposes and meet security restrictions. Diia will review applications within 10 working days.
Private contributors will be able to provide material voluntarily under licensing agreements that respect intellectual property rights. Access to a disputed dataset must be temporarily blocked within one working day of a complaint about improper use of a text, pending a review. After the platform launches, the digital ministry and education ministry, with participation from the National Academy of Sciences, will have six months to prepare proposals for expanding the national language-resource system.
Telegram · Follow KLYAMKA ↗